Plate Nº 57 · recorded November 17, 2026

Space & AstronomyReported finding

Satellite Plays Doom After First-Ever Voice Command From Orbit Test

ESA's PRETTY cubesat became the first spacecraft controlled by voice after Oslo radio amateurs spent 59 days and $775 to command it to play Doom from a rooftop.

By Marcus Bennett3 min read627 words

In brief

  1. July 27: first-ever voice command to a satellite, spoken from Oslo; PRETTY transcribed it on board and ran Doom.
  2. Radio amateurs spent 59 days and about 7,200 NOK ($775), pizza included, on the effort.
  3. Six failed attempts occurred over Poland and Norway between April and July before the seventh succeeded.
  4. Open-source tools GNU Radio and Sherpa-ONNX powered the signal processing and speech recognition.
  5. A July 3 test failed because the satellite's receiver channel was too wide; narrowing it fixed the problem.
Yes, it runs 'Doom': Satellite plays iconic game in world's 1st demonstration of orbital voice control
Plate Nº 57Yes, it runs 'Doom': Satellite plays iconic game in world's 1st demonstration of orbital voice control — WikiImages / pixabay

On July 27, a voice command spoken from a rooftop in Oslo reached the PRETTY cubesat in orbit, and the satellite — for the first time in history — understood it, transcribed it, and opened the 1993 video game Doom. The experiment, run by ESA's OPS-SAT Space Lab with software from Estonia's Tanagra Space, marks the first demonstration of voice control of a spacecraft.

The command itself was simple: "PRETTY play Doom Doom Doom," spoken at night by Jon Bergli Heier, call sign LB9BJ. PRETTY picked up the transmission, transcribed it using on-board voice recognition, and started a play session — then, like any gamer, filed a report. The satellite generated a "postcard" containing a frame of the gameplay, the transcription, and signal diagnostics, and retransmitted its recording of the command back to Earth.

Playing a first-person shooter was never in the design plan for PRETTY, a small satellite operated by a consortium that includes the European Space Agency. The real payload of the demo is the voice channel itself.

"Someone in distress at sea or in a remote area could transmit 'mayday' or 'help' on a radio," Tanagra Space said in a statement describing the possible future use. The long-term hope: a satellite equipped with transcription and AI could hear such a call and alert rescue services for people in remote zones.

Who actually pulled this off?

Not a professional ground station. The Oslo Group of the Norwegian Radio Relay League — hobbyists — made six unsuccessful attempts during satellite passes over Poland and Norway between April and July before the seventh try succeeded.

"We are radio amateurs, with the emphasis on amateurs. Satellite work was not something our club did, and we had no budget for it," group representative Eskil Hadland, call sign LB6AJ, said in the Tanagra statement. "So we borrowed a line wrongly credited to Pippi Longstocking: 'I have never tried that before, so I should definitely be able to do that.'"

The numbers behind the feat are modest. The Norwegians spent 59 days and roughly 7,200 NOK — about $775, pizza included — to reach PRETTY and help adapt the satellite's software to pick up a voice.

The debugging mattered as much as the shouting. A raw recording of a July 3 attempt, retransmitted to Earth, showed the spoken words were unrecognizable. Further work revealed the satellite's receiver channel was too wide to pick up the voice signal; narrowing the channel fixed it.

What made the transmission work?

The team leaned on open-source tools. GNU Radio handled signal processing, and Sherpa-ONNX handled speech recognition on board. That matters for practitioners: the entire voice pipeline runs on commodity, freely available software, not bespoke aerospace code.

The challenges were real. Voice-recognition software struggles with accents, background noise, and individual speaking styles on Earth; the satellite additionally contended with Doppler shift — the frequency distortion caused by its orbital motion — and an antenna never designed to relay human voices to space. The transmission worked anyway.

What could this change for spacecraft operators?

The team sees the demo as a stepping stone beyond spoken words. Future spacecraft could adapt the same approach to other radio frequencies and signal types. One suggested application: detecting deep-space radio signals with a satellite, with an on-board AI analyzing what it hears and sending operators only the most relevant data — filtering signal from noise autonomously.

These remain early-stage ideas, and the Doom session itself was a single demonstration, not an operational capability. No one played multiplayer from orbit; the satellite simply proved it could hear, parse, and act on a human voice.

For now, the Norwegians are enjoying the result. "Our tiny ground station shouted at an ESA spacecraft, and was heard," Hadland said.

via space.com (Source)

Filed under

  • voice-recognition
  • cubesat
  • satellite-communication
  • radio-amateurs
  • onboard-ai
Share this article:

More from Marcus Bennett

Marcus Bennett

Show full bio

News editor covering regulation, policy and their impact on general science news at SciBeat.

493 articles

Nearby plates

« Previous articleNext article »